NeurOWL: An LLM-Based Neural-symbolic Framework for Incomplete OWL Ontology Reasoning
Quick Answer
NeurOWL is a novel neuro-symbolic framework that addresses incomplete OWL ontology reasoning by integrating Large Language Models with ontology embeddings.
Quick Take
It effectively combines subsumption verification and ontology abduction, demonstrating robust performance across various real-world ontologies in domains like healthcare and bioinformatics.
Key Points
- NeurOWL unifies subsumption verification with ontology abduction for incomplete ontologies.
- The framework leverages both formal semantics and textual semantics.
- Evaluated on real-world ontologies, showing strong performance across multiple domains.
- Addresses challenges in reasoning due to incomplete real-world ontologies.
- Applicable in fields like healthcare and bioinformatics.
DeepSignal Analysis
What happened
NeurOWL is a new neuro-symbolic framework designed to tackle the challenges of incomplete OWL ontology reasoning. It integrates Large Language Models with ontology embeddings to perform subsumption verification and ontology abduction, addressing the issue of incomplete ontologies in various domains.
Key evidence
- NeurOWL focuses on a subsumption reasoning problem, determining the plausibility of a non-entailed subsumption in an incomplete ontology.
- The framework combines subsumption verification with ontology abduction, eliminating the need for a predefined set of missing axioms.
- Evaluation of NeurOWL on real-world ontologies in healthcare and bioinformatics shows strong and robust performance across different domains.
Why it matters
The integration of Large Language Models with formal ontology semantics represents a significant advancement in reasoning capabilities for incomplete ontologies. This could enhance the effectiveness of semantic reasoning in critical fields like healthcare and bioinformatics, where accurate knowledge representation is essential. By addressing the limitations of existing methods, NeurOWL may improve decision-making processes that rely on ontology-based reasoning.
Paper Resources
📖 Reader Mode
~2 min readAbstract:OWL ontologies provide a formal knowledge representation framework that enables semantic reasoning, and have been widely adopted across domains such as healthcare and bioinformatics. In practice, however, real-world ontologies are often incomplete, which pose challenges for reasoning. In this work, we focus on a fundamental subsumption reasoning problem: given an incomplete ontology and a candidate (non-entailed) subsumption, determine whether the subsumption is semantically plausible and, if so, providing a logically sound explanation containing potential missing axioms. This task unifies subsumption verification with ontology abduction, and generalizes the latter by removing the need for a predefined candidate set of missing axioms. To address this subsumption reasoning problem, we propose NeurOWL, an end-to-end neuro-symbolic framework that jointly performs verification and abduction, leveraging both formally defined semantics and textual semantics through Large Language Models and ontology embeddings. We evaluate NeurOWL on real-world ontologies across multiple domains, demonstrating strong and robust performance across different domains.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.15776 [cs.AI] |
| (or arXiv:2607.15776v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.15776 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hui Yang [view email]
[v1]
Fri, 17 Jul 2026 09:12:10 UTC (244 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →HOBA: Hierarchical On-Policy Bidding Agents for Adaptive Online Advertising
HOBA (Hierarchical On-policy Bidding Agents) is a novel hierarchical reinforcement learning framework that enhances online advertising bidding systems by improving adaptability and reducing hyperparameter tuning costs. It utilizes a for hyperparameter inference, a SARSA agent for expert model selection, and a dynamic expert pool for bid execution, achieving a +3.6% increase in target cost during large-scale deployment and outperforming state-of-the-art baselines on AuctionNet.